jiaxi2002

Upload folder using huggingface_hub

feb33a0 verified 4 months ago

5.36 kB

	import os
	import os.path as osp
	import base64 # 新增：导入base64模块
	from jinja2 import Environment, BaseLoader
	step = 4
	# 设置文件和文件夹路径
	caption_visualized_folders = [
	f"dataset/spotlight_sketch_cat/GT",
	f"dataset/spotlight_sketch_cat/epoch0"
	# f"test_1126_step10/epoch0",
	# f"test_1126_step25/epoch0",
	# f"test_1126_step{step}/epoch3",
	# f"test_1126_step{step}/epoch4"
	]

	# 提取模型名称
	model_names = ["GT", "sktech"]

	# 图片格式映射：后缀名 -> MIME类型（用于Base64的data URI）
	IMG_EXT_TO_MIME = {
	".jpg": "image/jpeg",
	".jpeg": "image/jpeg",
	".png": "image/png",
	".bmp": "image/bmp",
	".tiff": "image/tiff",
	".webp": "image/webp"
	}
	IMG_EXTS = set(IMG_EXT_TO_MIME.keys()) # 复用格式集合

	# 收集所有文件基名（stem）的并集，按字典序排序
	all_basenames = set()
	for folder in caption_visualized_folders:
	folder_local = osp.join(os.getcwd(), folder)
	if osp.isdir(folder_local):
	for fname in os.listdir(folder_local):
	ext = osp.splitext(fname)[1].lower()
	if ext in IMG_EXTS:
	all_basenames.add(osp.splitext(fname)[0])
	else:
	print(f"Warning: folder not found: {folder_local}")

	res_data = []
	for base in sorted(all_basenames):
	# 初始化前几个占位字段（模板主要使用索引后面的图片列）
	res_row_info = [base, base]

	# 对每个文件夹按基名查找对应的图片，编码为Base64
	for folder in caption_visualized_folders:
	base64_img = None # 存储Base64编码后的图片字符串
	img_filename = None # 存储图片文件名（用于显示）
	folder_local = osp.join(os.getcwd(), folder)

	if osp.isdir(folder_local):
	# 优先尝试常见扩展的有序匹配
	found = False
	for ext in [".jpg", ".jpeg", ".png", ".bmp", ".tiff", ".webp"]:
	candidate = osp.join(folder_local, base + ext)
	if osp.isfile(candidate):
	img_filename = base + ext
	# 读取图片并编码为Base64
	with open(candidate, "rb") as f:
	img_data = f.read()
	mime_type = IMG_EXT_TO_MIME[ext]
	# 拼接Base64的data URI格式
	base64_img = f"data:{mime_type};base64,{base64.b64encode(img_data).decode('utf-8')}"
	found = True
	break
	# 如果按扩展没有找到，尝试更宽松的匹配（忽略大小写后缀）
	if not found:
	for fname in sorted(os.listdir(folder_local)):
	fname_base, fname_ext = osp.splitext(fname)
	fname_ext_lower = fname_ext.lower()
	if fname_base == base and fname_ext_lower in IMG_EXTS:
	img_filename = fname
	candidate = osp.join(folder_local, fname)
	with open(candidate, "rb") as f:
	img_data = f.read()
	mime_type = IMG_EXT_TO_MIME[fname_ext_lower]
	base64_img = f"data:{mime_type};base64,{base64.b64encode(img_data).decode('utf-8')}"
	found = True
	break
	# 将Base64编码和文件名加入结果（无图则为None）
	res_row_info.append({
	"base64": base64_img,
	"filename": img_filename
	})

	res_data.append(res_row_info)

	# Jinja2模板（修改img的src为Base64编码）
	template_string = """
	<!DOCTYPE html>
	<html>
	<head>
	<title>Images</title>
	<meta charset="utf-8">
	<style>
	img { max-width: 320px; height: auto; display:block; margin:6px auto }
	td { vertical-align: top; text-align: center }
	table { width: 100%; border-collapse: collapse }
	th, td { border: 1px solid #ddd; padding: 8px }
	th { background: #f4f4f4 }
	.fname { font-size: 12px; color:#555; word-break:break-all; margin-top: 4px }
	</style>
	</head>
	<body>
	<h2>epoch0-1</h2>
	<table>
	<thead>
	<tr>
	<th>#</th>
	<th>GT</th>
	<th>sketch</th>
	</tr>
	</thead>
	<tbody>
	{% for row in data %}
	<tr>
	<td>{{ loop.index }}</td>
	{% for img_info in row[2:4] %} {# 要改 #}
	<td>
	{% if img_info.base64 %}
	<img src="{{ img_info.base64 }}" alt="img">
	<div class="fname">{{ img_info.filename or '未知文件' }}</div>
	{% else %}
	<div class="fname">(无图)</div>
	{% endif %}
	</td>
	{% endfor %}
	</tr>
	{% endfor %}
	</tbody>
	</table>
	</body>
	</html>
	"""

	# 渲染模板
	template = Environment(loader=BaseLoader).from_string(template_string)

	# 输出HTML文件
	output_html_path = f"showcase1203.html"
	with open(output_html_path, "w", encoding='utf-8') as f:
	f.write(template.render(data=res_data, model_names=model_names))

	print(f"Base64版HTML文件已保存为 {output_html_path}")